System Upgrade for the Adaptive Compression Point Control Algorithm on the CRE-331D at HZB

1. Project Context and Motivation

Large-scale particle accelerators rely on high-power Radio Frequency (RF) systems to accelerate, condition, and store particle beams. At the heart of these RF systems are Solid-State Amplifiers (SSA), which convert electrical AC power into a high-power RF signal. Today’s SSAs face a fundamental limitation: they are optimized for peak rated power, even though accelerators rarely operate under such conditions.

Instead, particle accelerators operate in different modes that require very different RF power levels, and for most of the time the demanded power is significantly lower than the maximum. Figure 1 illustrates typical operating modes and their corresponding RF power requirements at the BESSY II accelerator at HZB.

Figure 1: Logged RF power of the 4 amplifiers at BESSY II. The graph includes maintenance slots and unexpected outages.

When SSAs are operated below their rated output power, their efficiency drops noticeably. This means more electrical energy is drawn from the grid, leading to higher operating costs and unnecessary energy consumption.

Modern SSAs offer advanced control systems that allow the internal operating points of the internal RF transistors to be adjusted. In principle, this makes it possible to optimize efficiency for different power levels. In practice, however, these tuning options are rarely used. The reason lies in a gap between RF hardware expertise and accelerator operation: RF amplifier manufacturers know how to tune these parameters for their equipment, whereas accelerator operators focus on beam performance and reliability and cannot be expected to manually adjust low-level amplifier settings during routine operation.

To address this challenge, Cryoelectra developed the Adaptive Compression Point Control (ACPC) algorithm. The idea: instead of requiring detailed RF knowledge, the operator only specifies the required output power. The algorithm then automatically adjusts the amplifier’s internal settings to operate at the highest possible efficiency. In this way, ACPC translates high-level accelerator requirements into optimal low-level amplifier control.

Within the RF2.0 research consortium, Cryoelectra upgraded and tested an existing CRE-331D 500 MHz, 75 kW solid-state amplifier at HZB as a demonstrator. The goal was to show that significant efficiency gains can be achieved not only in new systems, but also by upgrading existing, already installed SSA legacy systems.

2. System Overview: The CRE-331D RF Power Amplifier

The CRE-331D is a 75 kW solid-state RF power amplifier operating at 500 MHz. Both the RF section and the original AC/DC power supplies were designed to support continuous operation at this maximum output power.

In real accelerator operation, however, the amplifier typically runs at much lower power levels. At HZB, for example, the average output power is around 35 kW, as shown in Figure 1. Nevertheless, occasional operating modes require higher power, which led to the amplifier being configured for up to 75 kW operation. Changing this maximum power setting was not foreseen during the original design and is not easily adjustable in daily operation.

A key parameter that determines both the maximum achievable RF output power and the efficiency at lower power levels is the supply voltage of the basic building block of the amplifier, the RF transistors. A higher drain voltage enables higher output power but results in lower efficiency when the amplifier is operated at reduced power. Conversely, lowering the drain voltage limits the maximum output power but significantly improves efficiency at lower output power.

In the original system, the available power supplies only allowed a drain voltage range between approximately 43 V and 50 V. This range is sufficient for full-power operation but too narrow to fully exploit efficiency improvements at lower power levels.

To enable adaptive and automated control, an upgrade kit was designed that includes a new generation of AC/DC power supplies with an output voltage range down to 23 V. This wider voltage range provides the necessary flexibility for the ACPC algorithm to dynamically optimize amplifier efficiency across all operating modes required by BESSY II.

3. Control System Upgrade and Algorithmic Approach

The ACPC algorithm runs locally on the solid-state amplifier control PC and is fully integrated into the existing control system. Its main design goal is simplicity.

From the user’s perspective, the algorithm has only a single input parameter: the required 1 dB compression point, which directly corresponds to the maximum usable and required RF output power. Based on this value, the algorithm automatically determines the optimal drain voltage for the RF transistors.

Before the algorithm can be used, a one-time characterization of the amplifier is required. This characterization links drain voltage, output power, and efficiency and serves as a calibration for the algorithm. For new systems, this step will be performed at the factory. For the CRE-331D demonstrator, the characterization was successfully carried out on site at HZB during commissioning of the demonstrator.

After the upgrade, the user interface remains almost unchanged. Only a single additional input field was added, allowing the operator to set the required output power. No manual tuning of internal amplifier parameters is needed anymore.

In addition to local control, the new parameter is also available via the remote control interface. This allows full integration of the upgraded amplifier into automated accelerator control systems and enables dynamic adjustment of the operating point depending on the accelerator mode.

4. Infrastructure Upgrade at HZB

The CRE-331D upgrade was designed as a retrofitting kit. Apart from the installation of the new AC/DC power supplies inside the amplifier system, no additional infrastructure changes were required at HZB. This demonstrates that the ACPC concept can be applied to existing installations with minimal effort, making it an attractive option for facilities aiming to improve energy efficiency without replacing complete RF systems.

5. Implementation Challenges and Project Timeline

During the initial installation phase, the project faced a minor setback. One crucial component of the retrofitting kit was missing and could not be sourced at short notice. As a result, the finalization of the demonstrator had to be postponed from December to mid-January.

Despite this delay, the issue was resolved without impacting the overall project goals, and installation and commissioning were completed successfully.

6. Current Status

The retrofitting kit is now fully installed, and the CRE-331D demonstrator at HZB is operational. The required amplifier characterization for the ACPC algorithm has been completed on site, and the upgraded SSA, including the new control algorithm, has been successfully tested.

The system is now ready for detailed performance evaluation under realistic operating conditions.

7. Measurement Results and Outlook

To estimate the impact of ACPC on accelerator operation and overall power consumption, real-life operational data were analyzed. Figure 1 shows the logged forward RF power of the four RF cavities at BESSY II during routine operation.

Based on this time-series data, the average wall-plug efficiency of the RF system was evaluated for four different cases:

  1. A constant 47 V drain voltage (Factory default).
  2. Operation with a constant 43 V drain voltage, representing the current best setting at BESSY II (Current setting).
  3. Post-retrofit operation with upgraded power supplies and a fixed drain voltage chosen according to the maximum required power. This corresponds to operation without ACPC (Post retrofitting).
  4. Full ACPC operation with optimal drain voltage selection for each operating mode (ACPC).

The results are summarized in Table 1. Shown are the effective wall plug efficiency and total reduction in AC power consumption relative to the current setting. The values in the “expected” columns are based on the extrapolation of the wall plug efficiency of the demonstrator to lower drain voltages before installing the retrofitting kit. The “measured” values correspond to the efficiencies measured during commissioning of the demonstrator including the new power supplies.

Each step leads to a significant increase in overall system efficiency. The largest improvement comes from the expanded drain voltage range enabled by the new power supplies. This upgrade alone reduces the AC power consumption by 10.6 %.

 Wall plug efficiencyAC power reduction
SSA settingexpectedmeasuredexpectedmeasured
Factory default35.4%35.4%-9.6%-9.6%
Current setting38.8%38.8%0.0%0.0%
Post retrofitting44.6%43.4%13.0%10.6%
ACPC47.2%45.1%17.8%14.0%

Table 1: Expected and measured average wall plug efficiency and reduction in required AC power for the typical operation of BESSY II.

The implementation of ACPC provides an additional efficiency gain. By selecting the optimal bias voltage for each operating mode, ACPC reduces AC power consumption by further 3.4 %pt.

There is a slight deviation in the expected and measured efficiencies for the cases Post retrofitting and ACPC. The measured efficiency is 1.2 %pt and 2.1 %pt lower than expected. This was traced back to the power supplies themselves showing a reduced AC to DC efficiency at lower output voltages. This partly counteracts the increasing efficiency of the RF subsystem at lower drain voltages. The overall effectiveness of the ACPC is therefore slightly lower than expected.

Still, the demonstrator project proved the validity of the approach. We were able to show that the ACPC can reduce the AC power consumption of the RF SSA at BESSY II by up to 14%.

Next steps include further characterization using additional real-life operation profiles from different accelerator projects to demonstrate the broad applicability of the concept. In parallel, additional testing will be carried out to verify the robustness of the upgraded system before integration into a particle accelerator.

Figure 2: The CRE-331D 500MHz 75kW amplifier during the demonstrator tests at HZB.

Figure 3: The ACDC-station of the demonstrator during rework.

Figure 4: Newly installed power supplies inside the CRE-331D amplifier.

In